On Tue, 15 Feb 2000, Brandon Metcalf wrote:
>
> Does anyone see a problem with using the following branching structure?
> First, the label LABEL_A may be at /main/LABEL_A or
> /main/branch_1/LABEL_A. Assuming LABEL_A is the baseline,
>
> element * CHECKEDOUT
> element * .../branch_2/LATEST
> element * LABEL_A -mkbranch branch_2
> element * /main/0 -mkbranch branch_2
>
> The above config spec would mean the branch_2 would originate from /main
> or from /main/branch_1. The alternative is,
>
> element * CHECKEDOUT
> element * .../branch_2/LATEST
> element * .../branch_1/LABEL_A -mkbranch branch_2
> element * .../branch_1/0 -mkbranch branch_2
> element * LABEL_A -mkbranch branch_1
> element * /main/0 -mkbranch branch_1
>
> This config spec would ensure that branch_2 always originates from
> /main/branch_1. Is this necessary, or would the first config spec cause
> any problems (merging, etc.) if branch_2 for one element originated from
> /main and for another element from /main/branch_1? Of course, in both
> examples, the baseline is the same.
I always use the first config spec. I indeed don't care where the branch
is sprouting from. For merges, I always use -fver '.../branch_2/LATEST'
combined with, for example, -element 'brtype(branch_1)' or whatever suits
your situation, and neither cares where the branch sprouts from.
For all intents and purporses, the branch tree may just as well be flat. I
have yet to see a use for differentiating between having a specific branch
type sprout from main or from somewhere else. The only two things I can
think of is a multisite situation where main is mastered by someone else
than branch_1, or a possessive QA team that doesn't trust clearcase and
that insists on having everything copy-merged into their own branch.
The multisite case either points out a real conflict, or just requires you
to use the correct resolution mechanism - perhaps certain merges into
branch_1 can only be performed on the site that I mastering branch_1, and
merges from branch_1 into branch_2 must be performed by the owner of
branch_2 - which is just an enforcement of the policy that made branch_1
be owned by a different site that branch_2.
By the way, multisite users out there... how do you decide who is to
master main?
-- cg- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
This archive was generated by hypermail 2b29 : Sun May 06 2001 - 00:23:15 EDT